DITA Session 4 Catch-up - Information Retrieval

I'm finally getting around to doing my information retrieval blog.

We had to conduct a number of searches using two different search engines: Google and Bing.  In addition to using those two search engines we had to do two different types of searches for each query.  In one search we used natural language queries (i.e. how to, where is, etc.) and in the other we had to use Boolean operators (i.e. NOT, AND, OR).  Then, we had to record our findings in an excel spreadsheet using the first 5 results and calculate the precision (number of relevant documents/number of documents retrieved).  We also had to label the querys by need type: transactional, navigational, information query.

I found that the search engines are better at or more precise at finding transaction queries.  I guess it's because what the user needs to do is more straightforward.  What I learned doing the first search in the activity is that these searches are really ambiguous because if you have limited knowledge of what you're looking for, it's harder to know if you've found it.  I thought that I'd found the correct website, but it turns out that I was way off base.  When I asked Andrew how was I supposed to know when I've found the right website, his response was that you'll know when you've found it.  Then I asked him what happens when you think you've got the right answer, but you're wrong.  He said that that is one of the things that makes information retrieval difficult.

I didn't find a big difference between using natural language queries or boolean operators.

It's much easier to do SQL searches if you know how to give the correct command.  It is much more straightforward because you know exactly what you want the system to produce.  Information retrieval is much more ambiguous because you don't always know what you need and you don't always know exactly what you're looking for.  Also, if you have limited knowledge of the subject of your search, you don't always know whether or not you've found it.  Additionally, the metadata, if any, that was attached to the information that you're looking for affects your search because if you're not using the right search terms, you may never find it.